Skip to main content

Table 2 Adjusted rate ratios of cerebrovascular outcomes associated with the use of warfarin and aspirin

From: A net clinical benefit analysis of warfarin and aspirin on stroke in patients with atrial fibrillation: a nested case–control study

  Cases/Controls Crude RR Adjusted RR (95 % CI)*
Ischemic stroke 5519/55,022   
No use of any therapy 1513/15,499 1.00 1.00 (reference)
Current use of warfarin monotherapy 896/13,238 0.67 0.65 (0.59, 0.71)
Below therapeutic range (INR: <2) 63/667 0.95 0.93 (0.71, 1.22)
Within therapeutic range (INR: 2–3) 132/1838 0.71 0.69 (0.57, 0.83)
Above therapeutic range (INR: >3) 31/361 0.86 0.82 (0.57, 1.20)
Unknown therapeutic range 670/10,372 0.64 0.62 (0.56, 0.69)
Current use of aspirin monotherapy 2002/18,399 1.11 1.05 (0.98, 1.13)
Intracranial hemorrhage 689/6858   
No use of any therapy 114/1365 1.00 1.00 (reference)
Current use of warfarin monotherapy 242/2214 1.41 1.29 (1.00, 1.68)
Below therapeutic range (INR: <2) 13/126 1.32 1.16 (0.62, 2.16)
Within therapeutic range (INR: 2–3) 34/356 1.25 1.13 (0.74, 1.72)
Above therapeutic range (INR: >3) 13/47 3.63 3.26 (1.67, 6.38)
Unknown therapeutic range 182/1685 1.39 1.29 (0.98, 1.69)
Current use of aspirin monotherapy 172/2210 0.97 0.92 (0.70, 1.19)
  1. Abbreviations: RR: Rate ratio; CI: Confidence interval; INR: International normalized ratio.
  2. *Adjusted for excessive alcohol use, smoking status, obesity, peripheral artery disease, myocardial infarction, previous cancer, prior bleeds, thromboembolic disorders, ACE inhibitor use, angiotensin receptor blocker use, antidepressant use, antipsychotic use, NSAID use, and statin use. The stroke model was additionally adjusted for CHADS2 score, while the intracranial hemorrhage model was additionally adjusted for the components of that score.
  3. Mutually exclusive categories among current users of warfarin monotherapy.
  4. Note: Current users of other antithrombotic therapies or combinations, as well as past users are not displayed in the Table, but were included in the regression model to allow the proper estimation of treatment effects. These represent 1108 cases and 7886 controls for the ischemic stroke model, and 161 cases and 1069 controls for the intracranial hemorrhage model.